By accounting for spatial variation in soil N levels, variable-rate fertilizer application may improve crop yield and quality, and N use efficiency within fields. The main purpose of this study was to demonstrate how site-specific wheat yield and protein data, and a geographic information system may be used in developing precision N-recommendations for spring wheat. The three steps in the procedure include: (1) estimate the amount of N-removed in wheat in the year in which the crop is harvested, (2) estimate the N-deficit, defined as the amount of additional N needed for raising protein concentration in a future crop to a specified target level, and (3) estimate the total N-recommendation by summing the mapped values of the N-removed and the N-deficit. A map for variable-rate application of fertilizer is derived by specifying cutoff values to divide the range in the total N-recommendation into classes representing N management zones. A field experiment was conducted within an annually cropped wheat field 101 ha in northern Montana to determine whether the proposed method could improve grain yields and protein levels. The N-removal and N-deficit were estimated from site-specific wheat yield and protein data that were acquired during harvest of 1996. In 1997, which was a dry year, an experiment was conducted in the same field that consisted of a randomized complete block design arranged as pairs of strip plots. Variable- or uniform-rate N treatments were randomly assigned to each pair of strips. Both treatments received nearly the same amount of fertilizer, however, N in the variable treatment was varied to match patterns in grain yield and protein levels that previously existed in 1996. Yields were not significantly different between management systems, but proteins were significantly enhanced by spatially variable N application. In addition, variability in protein levels was reduced within the whole field. Field areas deficient in N fertility could be identified without having to sample for soil profile N.
